Insurance companies are all about managing risk. But to succeed, they must also get products to market quickly, while reducing costs.
And it’s almost inevitable that they will have to bring legacy technology like mainframes along for the ride.
Because when it comes to the compute power needed to underpin 200-year US insurance firm The Hartford, there really isn’t an alternative, the firm’s head of IAC, Automation, PaaS, Middleware, VDI, AJ Oller explained at HashiConf in Boston this week.
“At the core of all of our systems, your big banks, your insurance companies, we all live with a mainframe and trying to automate and modernize around it really does bring a lot of challenges.”
The most recent transformation overseen by Oller has involved moving all of its applications to cloud-based infrastructure automation.
The firm had already adopted a cloud-like approach with its on-prem datacentre operations using Terraform Enterprise, Oller said.
It worked closely with HashiCorp specialists River Point Technology to plan the migration to HCP Terraform, he said. Throughout, the aim was to ensure that internal staff would be ultimately able to run systems themselves in the future.
A key element of this was pair programming “So that by the end of it, our team knows how to fish, and a lot of it is [Riverpoint’s] team bringing expertise, injecting themselves into what we're doing from a design and implementation perspective, but we actually do the implementation”
One key benefit of this, he said, was that the if there were any “stragglers and workspaces or any customers that had to stay behind on TFE a little bit longer, for whatever reason, we knew how to do this migration on our own afterwards.”
Similarly, he says, the team adopted a platform approach early on rather than having the entire development team learn to write their own Terraform modules.
“So, what we try and do is look at an application holistically, look at the components that need to be built out and deployed, and for the most part, these things are already in my shop.”
That can mean anything from integration components, to databases, to load balancers, Oller said. “And then we say, ‘Well, how can we right size this application?’ Once we have their IAC, essentially, we hand it over, and that becomes their day zero.”
Ultimately, he said, the migration project, start to finish, took less than a month. The actual migration of thousands of workspaces was carried out in 24 hours. He described it as 99.99 percent successful. “With the exception of one team that used the CLI in production. It would have been pretty hard to identify who they were because they weren't using any of our CI/CD pipelines. So, you do have some outliers that might fall through the cracks,”
One key benefit from rolling out Terraform and ramping up automation comes in patch management, Oller said. “We have a self-renewing infrastructure footprint where folks are rebuilding every 90 days. We're talking about bringing that to every 60 days.” So, when it comes to audit and security requirements, “We've shifted all the way left.”
New versions of Terraform Modules were typically delivered every month, he continued, though things are typically expedited in the case of zero days or other vulnerabilities. “But the result right is we have cleaner machines and a whole lot less overhead in managing audit and security practices.”
And as for the mainframe? As Oller points out it’s rare for new applications or capabilities to be developed on mainframes. But ultimately, the big iron is not going anywhere. Cloud or colocation options notwithstanding.
Afterall, he told The Stack, “We have over 200 years of historical rating data and insurance data that is pretty hard to compute outside of the mainframe...I think its sheer computing power is the reason you see it have such staying power throughout so many industries.”
The challenge is not to be completely bound by the legacy, he says. “When you're 200 plus years old, it really, really helps so that you're not necessarily bringing a legacy mindset into modern toolsets.”
At the same time, he said, that legacy also means “You have a wealth of expertise, of folks that have seen several infrastructure or digital transformations happen.” And that’s not something that should be treated lightly. “You can bring some of these lessons learned along the way into your next transformations.”
